The Stability of Same-Sex

Cohabitation, Different-Sex
Cohabitation, and Marriage

Charles Q. Strohm

PWP-CCPR-2010-013

Latest Revised: February 1, 2012

California Center for Population Research

On-Line Working Paper Series

1
Running head: Stability of same-sex cohabitation

DRAFT: DO NOT CITE WITHOUT AUTHORS PERMISSION

The Stability of Same-Sex Cohabitation, Different-Sex Cohabitation, and Marriage

Charles Q. Strohm
Research Triangle Institute
California Center for Population Research
* This research was funded by the National Science Foundations Graduate Research Fellowship
Program and by the Institute for Gay and Lesbian Strategic Studies. I thank Judith Seltzer, Kate
Choi, Robert Mare, Megan Sweeney, Susan Cochran, Gary Gates, Holning Lau, and members of
the UCLA Demography, Family, and Stratification Research Group for valuable insights. I am
grateful to Peter Shepherd and Brian Dodgeon at the Centre for Longitudinal Studies for
assistance with the NCDS and BCS data. Any errors are my own. Earlier versions of this paper
were presented at the 2008 annual meetings of the Population Association of America and the
American Sociological Association.

Abstract
This study contributes to the emerging demographic literature on same-sex couples by
comparing the level and correlates of union stability among four types of couples: male same-sex
cohabitation, female same-sex cohabitation, different-sex cohabitation, and different-sex
marriage. I analyze data from two British birth cohort studies, the National Child Development
Study (N = 11,469) and the 1970 British Cohort Study (N = 11,924). These data contain
retrospective histories of same-sex and different-sex unions throughout young adulthood (age
16-34) from 1974-2004. Event history analyses show that same-sex cohabitations have higher
rates of dissolution than do different-sex cohabiting and marital unions. Among same-sex
couples, male couples had slightly higher dissolution rates than did female couples. In addition,
same-sex couples from the 1958 and 1970 birth cohorts had similar levels of union stability. The
demographic correlates of union stability are generally similar for same-sex and different-sex
unions.

Keywords: Cohabitation; Gay, Lesbian, Bisexual, Transgender; Marriage; Social Support; Social
Trends / Social Change; Stability

3
In an era of high divorce rates, family scholars have sought to identify the forces that
contribute to the stability of couple relationships. In doing so, researchers have compared the
stability of marriages and unmarried cohabitations, marriages preceded by cohabitation and
marriages that were not, and unions comprised of partners with varying demographic
characteristics. Although less studied than other union types, same-sex couples also represent a
unique opportunity to study cohesion in couple relationships. Most same-sex couples currently
lack the institutionalization that underlies couple stability, but jurisdictions increasingly offer
legal recognition to same-sex couples. Although same-sex couples remain socially stigmatized,
attitudes are also growing more accepting. These shifts allow researchers to study how legal and
social institutionalization contributes to the stability of couple relationships (Biblarz & Savci,
2010). In addition, the gender composition of same-sex couplestwo men or two women
allows researchers to study how gender affects the stability and functioning of intimate
relationships more generally (Blumstein & Schwartz, 1983).
An important first step in investigating these theoretically rich issues is to describe the
demography of same-sex unions. Spurred by the growing availability of data, researchers have
begun to study the stability of same-sex unions in Sweden and Norway (Andersson, Noack,
Seierstad, & Weedon-Fekjaer, 2006) and the Netherlands (Kalmijn, Loeve, & Manting, 2007).
My research builds on these studies by studying a new set of correlates (e.g., birth cohort) and
describing the stability of same-sex couples in Britain, a country with less legal and social
acceptance of same-sex couples compared to the Netherlands, Norway, and Sweden (European
Commission, 2007). In this paper, I compare the levels and correlates of stability for male and
female same-sex cohabitation, different-sex cohabitation, and different-sex marriage (marriage
for brevity). I also explore whether the correlates of union stabilityincluding birth cohort,

4
union history, and family backgroundare similar for the different types of couples. To
investigate these questions, I analyze data from two British cohort studies, the National Child
Development Study (NCDS; 1958 birth cohort) and the 1970 British Cohort Study (BCS; 1970
birth cohort). The NCDS and BCS collected retrospective histories of same-sex and different-sex
coresident unions since age 16 (Bynner, Butler, Ferri, Shepherd, & Smith, 2005). Because these
data focus on couples that live together, I refer to coresident unions as couples in this paper for
brevity. The NCDS and BCS are unique in that they contain comparable histories of same-sex
and different-sex unions for two young adult cohorts spanning a long time period (1974-2004).

Background
Levingers theory of marital cohesion (Levinger, 1965, 1976) offers clues into the levels
and correlates of stability for same-sex and different-sex couples. Grounded in social exchange
theory, Levingers theory posits that union stability is a function of three factors: rewards from
the relationship such as emotional support and social prestige; barriers to leaving the
relationship such as legal requirements and joint investments; and attractive alternatives to the
relationship such as another partner or being single. In this section, I discuss how the legal and
normative context influences same-sex couples perceptions of the barriers, rewards, and
alternatives to their unions, giving rise to differences in union stability. I begin by comparing the
levels of stability between same-sex and different-sex unions. Next, I compare the levels of
stability among male and female same-sex unions. I conclude by exploring the correlates of
stability among same-sex and different-sex unions.

The Stability of Same-Sex and Different-Sex Unions

5
Same-sex couples experience a different legal and normative climate than do differentsex couples. Marriage between same-sex partners is not legally recognized in most parts of the
world. And despite growing acceptance of same-sex relationships, same-sex couples continue to
be socially stigmatized (Meyer 2003). The legal and normative climate means that same-sex
couples likely perceive fewer barriers to exit, fewer rewards from the relationship, and greater
alternatives compared to different-sex couplesleading to lower rates of stability.
Consider, first, the barriers to dissolution faced by same-sex couples. The lack of samesex marriage means that same-sex couples rarely face legal hurdles when they seek to dissolve
their relationship. Although legal barriers may themselves be insufficient to preserve an unhappy
marriage (Knoester & Booth, 2000), the cost and time required for divorce may lead some
different-sex couples to reconsider a divorce or resolve their problems (Preveti & Amato, 2003).
In addition to legal hurdles, marriage also creates other barriers to exit, including relationshipspecific investments such as children, specializing in paid and unpaid work, and pooling income
(England & Kilbourne, 1990). These investments in the relationship foster interdependence and
provide incentives to remain in the relationship (Treas, 1993). Marriage encourages relationshipspecific investments because it reduces uncertainty about the relationships future and provides
insurance against the risks of investment (Brines & Joyner, 1999; England & Kilbourne, 1990).
Same-sex couples may hesitate to make relationship-specific investments for several
reasons. First, the inability to marry creates uncertainty about the relationships future and
partners responsibilities to each other. Same-sex couples also lack legal protections for their
investments such as marital property rights (Herek, 2006). Consistent with this perspective,
same-sex couples are less likely than different-sex married couples to pool economic resources
(Solomon, Rothblum, & Balsam, 2004) and specialize in paid and unpaid work (Black, Sanders,

6
& Taylor, 2007). Similarly, different-sex cohabiting couples make fewer relationship-specific
investments than do married couples (Haskey, 2001; Heimdal & Houseknecht, 2003).
Second, same-sex couples may make fewer relationship-specific investments because of
the social stigma associated with homosexuality. Lehmiller and Agnews research on social
marginalization shows that members of marginalized couples invest less in the relationship than
do those in non-marginalized relationships (Lehmiller & Agnew, 2006; Lehmiller & Agnew,
2007). Marginalized couples may prefer to keep their relationship secret or less visible to avoid
unpleasant experiences with disapproving friends, family, or strangers. Lehmiller and Agnew
(2006) argue that it is difficult to make relationship-specific investments while simultaneously
maintaining a low visibility for ones relationship. Therefore, same-sex couples, who face
disapproval of their relationship, might hesitate to invest in their relationship by establishing ties
with their partners family (Solomon et al., 2004) or by moving in together (Strohm, Seltzer,
Cochran, & Mays, 2009) compared to both different-sex married and unmarried couples, who do
not experience the stigma of homosexuality.
In sum, with fewer investments to bind them together, same-sex couples face fewer
barriers to exiting the union than different-sex couples, particularly different-sex married
couples. But in addition to barriers, union stability also depends on the availability of attractive
alternatives such as being in another relationship or being single. Same-sex couples may have a
greater number of alternatives to the relationship compared to different-sex (married and
unmarried) couples. Although the population searching for a same-sex partner is relatively small,
many lesbians and gay men are geographically clustered in some urban settings (Black, Gates,
Sanders, & Taylor, 2000; Ellingson & Schroeder, 2004), providing a large pool of alternative
partners and translating into less stability. Further, singlehood may be a more attractive status for

7
lesbians and gay men compared to heterosexuals. Many sexual orientation minorities, in
response to anti-gay stigma, have developed thriving friendship networks, sometimes referred to
as families of choice (Muraco, 2006; Weston, 1997). The salience of friendship means that
there are fewer costs to relationship dissolution and becoming single for lesbians and gay men
compared to heterosexuals.
Same-sex couples may perceive fewer rewards for being in their relationship compared to
different-sex married couples. This is not because same-sex partners provide each other fewer
rewards such as emotional support: same-sex and different-sex couples report being similarly
satisfied with their relationships (Kurdek, 1998). Rather, same-sex couples likely derive fewer
social rewards for being in their relationship. For different-sex couples, marriage is a sign of
social prestige and marks a successful transition to adulthood (Cherlin, 2004). Most same-sex
couples, however, cannot enjoy the social recognition of marriage nor the legitimacy of the
relationship that marriage provides. Indeed, same-sex couples perceive less support for their
relationship from their family compared to their married counterparts (Kurdek, 2004).
In sum, the lack of marriage and normative support means that same-sex couples perceive
fewer barriers to leaving the union and more alternatives to the relationship compared to both
different-sex married and unmarried couples. Further, same-sex couples are likely to perceive
fewer rewards compared to different-sex married couples. Because barriers, rewards, and
alternatives are associated with union dissolution, it follows that same-sex couples will
experience greater levels of instability than different-sex couples, particularly married couples.
Previous research supports these predictions (Balsam, Beauchaine, Rothblum, & Solomon, 2008;
Blumstein & Schwartz, 1983; Kurdek, 2004). In particular, two studies use population-based,
longitudinal data to study the stability of same-sex unions. In Sweden, the dissolution rate for

8
male and female same-sex marriages was 1.4 and 3.0 times, respectively, greater than the rate
for different-sex marriages (Andersson et al., 2006). This difference persisted even when
analysis was restricted to childless couples, suggesting that same-sex couples lower likelihood
of having children does not fully explain their greater instability. Kalmijn et al. (2007) compared
same-sex cohabiters, different-sex cohabiters, and different-sex married couples in the
Netherlands between 1989 and 1999. In these longitudinal data, different-sex cohabiters who
later marry are first classified as cohabiters and are then subsequently classified as married. The
dissolution rate for same-sex cohabitation was twelve times higher than the rate for different-sex
marriage, and three times higher than the rate for different-sex cohabitation (Kalmijn et al.,
2007).

The Stability of Male and Female Same-Sex Unions

Few studies have examined the relative stability of male and female same-sex unions
(Blumstein & Schwartz 1983; Kurdek, 2003). Differences in the social psychology and economic
circumstances of men and women, however, suggest that women may experience lower
dissolution rates compared to men. For example, the social psychological theory of self-construal
(Cross & Madson, 1997) posits that women are more likely than men to adopt an interdependent
understanding of the self, or self-construal, that is rooted in relationships. Mens self-construal,
in contrast, is more likely to be grounded in autonomy and individualistic pursuits (Guimond,
Chatard, Martinot, Crisp, & Redersdorff, 2006). Thus, self-construal theory suggests that women
in same-sex couples may be more likely than men to derive meaning through their relationship
(Kurdek, 2003) and perceive more rewards from being in a relationship. Women may also
perceive greater rewards than men because of economic concerns: Womens lower wages may

9
lead women to view the economies of scale provided by cohabiting unions more favorably than
do men. In addition to greater rewards, women may also perceive more barriers to exiting samesex unions because female same-sex couples are more likely to have children than are male
same-sex couples (Black et al., 2007). Finally, men might also have more alternatives to a
relationship because men are more likely than women to live in large urban areas (Black et al.,
2000) that sometimes contain highly organized social networks (Ellingson & Schroeder, 2004).
In sum, female couples may be more stable than male couples because women perceive
more rewards, more barriers, and fewer alternatives than men. But these very factors may lead to
greater stability among male same-sex cohabiters because of the different types of women and
men who chose to enter same-sex cohabiting unions. For example, if women are particularly
motivated to enter same-sex unions for social psychological or economic reasons, then a broad
range of female couples (some highly committed, others less committed) may progress to
coresident relationships. In contrast, only the most committed male couples may move in
together because they perceive fewer rewards from relationships. Because commitment affects
stability (Kurdek, 1995), it follows that the more highly selective group of male couples may
be more likely to remain intact compared to the heterogeneous group of female couples.
The mixed empirical record on the stability of male and female couples reflects these
divergent predictions. For example, while Kalmijn et al. (2007) find greater stability among
female same-sex cohabiters compared to male cohabiters in the Netherlands, Andersson et al.
(2006) find that male same-sex married couples in Norway and Sweden are more stable than
their female counterparts. The available empirical evidence is inconclusive, in part because of
the small number of studies, the varying definitions of couples, and the different contexts. By

10
comparing the stability of same-sex and different-sex cohabiting unions in Britain, this paper
provides additional information to help resolve the mixed empirical record.

The Correlates of Union Stability

There is little reason to anticipate differences between same-sex and different-sex couples
for some well-known correlates of stability such as childhood family structure or socioeconomic
status (Berrington & Diamond, 1999), occupation and school enrollment (Ermisch &
Francesconi, 2000), and childhood region (Berrington & Diamond, 1999). These correlates are
likely to affect same-sex and different-sex couples similarly. For example, entering a union at a
young age is positively associated with dissolution for different-sex couples (Ermisch and
Francesconi, 2000) due to emotional immaturity or the poor quality of matches early in life.
These mechanisms are likely to apply to same-sex couples. Indeed, Andersson et al. (2006)
report a negative association between age of entry and the dissolution of same-sex marriages.
But some correlates might have different associations for same-sex and different-sex
unions. Birth cohort is one example. There are unlikely to be differences between the 1958 and
1970 cohorts in the stability of different-sex married (Wilson & Smallwood, 2008) or differentsex cohabiting (Berrington, 2003) couples. However, the context of same-sex partnering was
markedly different for the 1958 and 1970 birth cohorts. Public opinion had become more tolerant
of homosexuality: 86% of British adults said homosexuality was always, mostly, or
sometimes wrong when the 1958 cohort was 29 years old (1987), compared to 62% when the
1970 cohort was 29 years old (1999) (British Social Attitudes Surveys; authors computations).
The legal context had also changed. Whereas the 1958 cohort came of age during a resurgence of

11
anti-gay policies (Jivani, 1997), the 1970 cohorts young adulthood was marked by a series of
policy victories for lesbians and gay men (Tobin, 2009).
The increasingly tolerant normative and legal climate leads to two opposing predictions.
On one hand, young adults from the 1970 cohort may have had more supportive families and
encountered less stigma, leading to greater stability for the 1970 cohort. On the other hand,
selection processes may lead to the opposite pattern. The more unfavorable social context of
the 1958 cohorts young adulthood may mean that only the most committed couples from that
cohort would move in together. In contrast, a more heterogeneous group of couples from the
1970 cohort may have moved in together, suggesting greater stability among the 1958 cohort.
Like birth cohort, union history may operate differently for same-sex and different-sex
couples. For different-sex couples, having previously cohabited or married is associated with
dissolution (Steele, Kallis, Goldstein, & Joshi, 2005), in part because it signals poor relationship
skills or liberal attitudes toward dissolution (Steele, Kallis, & Joshi, 2006). For same-sex
couples, union history could also be a proxy for these risk factors. But a significant proportion of
same-sex cohabiters have entered a different-sex union earlier in life (Andersson et al., 2006;
Black et al., 2000) due to normative pressures to enter a different-sex union or changes in partner
preferences. Because same-sex cohabiters might have experienced dissolution for reasons other
than having poor relationship skills or liberal attitudes, union history might be a less robust
predictor of dissolution for same-sex couples than for those in different-sex relationships.

Method
Data

12
I pool data from the National Child Development Study (NCDS) and the 1970 British
Cohort Study (BCS) (Bynner et al., 2005). The NCDS and BCS are prospective cohort studies of
all people born in Britain in a particular week in 1958 and 1970, respectively. Children who
immigrated to Britain by age 16 were also added to both cohorts. The NCDS and BCS have
collected information from cohort members and their families periodically since birth. For the
NCDS, I use data from the 11,469 individuals who participated in the paper-and-pencil age 33
interview, which represents 71% of eligible cohort members. I use data from the 11,924 BCS
cohort members who participated in the age 30 or age 34 computer-assisted personal interview
(CAPI) (74% of eligible cohort members). The NCDS and BCS remain largely representative of
the original cohort, though there was more attrition among men and socioeconomically
disadvantaged individuals (Hawkes & Plewis, 2006).
I analyze data from retrospective histories of coresident unions collected during in-person
interviews at age 33 in the NCDS and at age 30 and 34 in the BCS. The NCDS and BCS did not
collect complete information about nonresident relationships or periods of living apart from
cohabiting partners or spouses. Because the NCDS and BCS were designed to be comparable,
the question wording was nearly identical across surveys. For each coresident union that lasted
one month or more since age 16, cohort members reported the month and year they began living
with each partner, the month and year of marriage (if applicable), and whether the union ended
through dissolution or a partners death (if applicable). A unique feature of the NCDS and BCS
is that cohort members were asked to report the sex of each previous partner (Was this person
male or female?), allowing me to classify previous unions as same-sex or different-sex. I
classify current cohabitations as same-sex or different-sex by linking the cohort members sex
with the sex of the cohort members partner from the household roster.

13
After excluding 447 cohort members (1.9% of both cohorts) who provided incomplete or
inconsistent union histories, there were 20,070 cohort members who reported at least one
coresident union. Those who grew up in a non-traditional family, people from socioeconomically
advantaged backgrounds, and those in low status occupations in young adulthood were more
likely to have inconsistent or incomplete union histories. I include individuals first and higherorder unions in the analysis and control for whether the respondent had a previous union in the
analysis. There were 186 cohort members who ever entered a same-sex cohabitation; these
cohort members reported a total of 263 same-sex relationships (138 from men and 125 from
women). There were 17,219 unions (8,363 from men, 8,856 from women) that began as
different-sex cohabitations, of which 8,663 eventually transitioned to marriage. Finally, there
were 8,174 different-sex unions that began as marriages with no premarital cohabitation (3,641
from men, 4,533 from women). Using this information, I created a person-month file in which
each partnered individual contributes one record for each month the union is at risk of
dissolution (2,136,313 total person-months). Unions are at risk of dissolution until the
relationship ends (for marriages, I use the date of separation rather than divorce) or the union is
right-censored. Right-censoring may occur due to the partners death, attrition, or if the union
remains intact at the end of the observation period. Because individuals may marry their
different-sex cohabiting partner, I treat union type as time-varying: cohabiting individuals who
marry their partner are first classified as cohabiting and are then classified as married. Following
Kalmijn et al. (2007), the variable measuring the relationships duration continues to increase
after the relationship transitions to marriage; this clock is not reset once they partners marry.
There are two main challenges to measuring same-sex relationships in large-scale
surveys. First, like other data on sensitive subjects, the data I use may suffer from social

14
desirability bias: Individuals in same-sex unions may not agree to be interviewed to maintain
privacy about their relationship. Alternatively, individuals who are interviewed may simply omit
same-sex unions from their reports, particularly in interviewer-administered surveys such as the
NCDS and BCS. I consider possible social desirability biases in my interpretation of the results.
It is worth noting that the percentage of unions that are same-sex in the NCDS and BCS data is
similar to other data sources. In the NCDS/BCS samples, same-sex couples account for 1.0% of
all couples. Andersson et al. (2006) report that same-sex couples accounted for 0.7% of all
couples in Swedish marriage register data between 1993-2001. Black et al. (2000), who use U.S.
Decennial Census data, report that same-sex couples account for 0.2% of all couples.
A second potential problem is classification error, whereby the sex of a respondents
partner is miscoded, leading to a different-sex union being misclassified as a same-sex union (or
vice versa). This problem has been documented in paper-and-pencil modes such as in the U.S.
Decennial Census, in which stray marks or respondent error can lead to misclassification (Black
et al., 2000). Even a small rate of misclassification of different-sex couples can result in
significant contamination of the smaller group of same-sex couples (Black et al., 2000).
Misclassification of partners sex may be present in the paper-and-pencil NCDS, but is unlikely
to occur in computer-assisted surveys such as the BCS because of additional quality controls
(Gates & Steinberger, 2010). Fortunately, the age 33 NCDS interview collected union histories
twice: once during a self-completion questionnaire and another time during the face-to-face
CAPI-assisted interview. I used this unique information to reduce measurement error of samesex unions. In a supplementary analysis (results not shown), I examined the consistency between
the two data sources. There were substantial inconsistencies: of all same-sex unions, only 22%
were reported in both sources. In this paper, I defined a union as same-sex if it met the following

15
conditions: (1) the respondent reported the union being same-sex in at least one data source, (2)
the respondent never reported marrying their partner, (3) for current unions, the partners sex in
the household roster matched the partners sex in the union history. This resulted in 86 same-sex
unions from the NCDS. Another group of researchers analyzed these data and created an
alternative definition of same-sex unions using slightly different assumptions (Di Salvo, 1995).
These researchers reported 76 same-sex unions in the NCDS. In a supplementary analysis, I
found that using this alternative classification scheme did not appreciably affect the parameters
in my models.

Independent variables
In addition to gender and birth cohort, I include mothers education, childhood family
structure, and childhood region of residence. In preliminary analyses, I experimented with
different measures of childhood socioeconomic status such as fathers education and fathers
social class. Mothers education had the closest fit to the data (results not shown). I define cohort
members who were born to an unmarried mother or whose parents separated or divorced as
growing up in a non-traditional family. I identify cohort members who have been married or
lived with a partner previously using a time-varying, dichotomous variable. I use a linear and
squared term for the cohort members age of entering the union. A time-varying variable, lagged
by one month, is used to identify respondents enrolled in full-time education. Socioeconomic
status is a time-varying indicator of occupational and employment status, lagged by one month
with the following categories: high-skilled (e.g., managerial, professional), medium-skilled (e.g.,
clerical, sales), low-skilled (e.g., machine operation), not working (e.g., unemployed, retired),
and missing (Gregory, Zissimos, & Greenhalgh, 2001). Occupational skill level is a commonly-

16
used measure of socioeconomic status in Britain. Using a six or twelve month lag rather than a
one month lag for this variable did not change the results. I treat missing data as an additional
category of these independent variables. I do not consider children because there were only 14
NCDS cohort members and 47 BCS cohort members who had ever entered same-sex unions and
also had children by age 34less than 1% of the samples. In contrast, 75% of marriages and
24% of cohabitations in the NCDS had children (Steele et al. 2005). Other than the partners sex,
the NCDS and BCS do not contain information about previous partners characteristics.

Analysis
I begin by describing the characteristics of individuals in each union type and showing
the probabilities of union dissolution using life tables. Next, I conduct a series of discrete-time
event history models predicting whether a union dissolves in a particular month. The event
history analysis proceeds in two stages. The first stage tests for differences in the levels of
stability across the four union types (female same-sex cohabitation, male same-sex cohabitation,
different-sex cohabitation, and marriage). In this stage, I estimate a discrete-time logistic
regression of dissolution on duration, union type, cohort, and family background (childhood
region, childhood family structure, and mothers education). Next, I estimate two sequential
models in which I first add age of union entry and union history to the model, and then add
education enrollment and occupation/employment. These sequential models illustrate how
compositional factors may explain differences in the stability of same-sex and different-sex
unions. All models are adjusted for the clustering of unions within individuals. In all models,
duration of the relationship is represented by two monthly linear splines: years 1-2 and year 3.

17
This specification had the closest fit to the data according to the BIC criteria (results not shown).
All analyses are unweighted and were conducted using Stata version 10.
The second stage investigates the correlates of dissolution for each union type. In this
stage, I estimate the final model from the first stage of the analysis, but stratify by union type
(same-sex cohabitation, different-sex cohabitation, marriage). Due to small sample sizes, I pool
male and female same-sex cohabitation. I use these models to test for the significance of
variables within a union type (e.g., does the odds ratio for cohort differ from 1.0), as well as to
test for differences of variables across models (e.g., does the odds ratio for cohort differ between
same-sex and different-sex cohabitation). In a supplementary analysis reported in the text, I
included interaction terms between occupation and gender to investigate whether the associations
between occupation and dissolution differ by gender.

Results
Characteristics of individuals and their unions
Panel A in Table 1 shows the demographic characteristics of individuals who report ever
entering same-sex cohabitation, different-sex cohabitation, and marriage. Because I include an
individuals first and subsequent unions, the columns are not mutually exclusive: An individual
who cohabited with a different-sex partner and then cohabited with a same-sex partner would be
represented in both the first and second columns. The results show that both same-sex and
different-sex cohabiters were more likely than are married individuals to be from the 1970 cohort
and live in a non-traditional family during childhood. Individuals who ever cohabited with a
same-sex partner were more likely to be in full-time education at age 23 and have higher
occupational attainment at age 29, relative to those who ever enter different-sex cohabitation or

18
marriage. The relatively high socioeconomic status of same-sex cohabiters is not due to
differences in family background: The three groups did not differ in mothers education.
[Table 1]
The bottom panel of Table 1 describes the characteristics of unions. Forty percent of
same-sex cohabitations were preceded by another union, compared to only 29% of different-sex
cohabitations and 5% of marriages. In other words, same-sex cohabitations were more likely to
be an individuals second or higher-order union compared to marriage and different-sex
cohabitation. The median age of entry into same-sex cohabitation was 25.9 years. This is higher
than the median age of entering different sex cohabitation (24.6 years) or transitioning from
single to marriage (24.5 years) (p < .05). In a supplementary analysis of same-sex cohabitation,
there was no difference between men and women in the number of previous unions or the
median age of union entry (results not shown).

Life table estimates of dissolution

In Table 2, I show the probabilities that each union type will remain intact for the first
eight years of the union. As in Table 1, individuals can contribute multiple relationships in this
analysis. For the purposes of this analysis only, I use the union type at the beginning of the
relationship; cohabitations that transition to marriage are treated as cohabitations. Subsequent
analyses treat union type as a time-varying variable. The results show that marriage is the most
stable union type, followed by different-sex cohabitation, and then by same-sex cohabitation. For
example, the probability of a union lasting five years was .88 for marriage, .67 for different-sex
cohabitation, and .37 for same-sex cohabitation. This survival probability for marriage is similar
to that reported in Britain using vital records data (Wilson & Smallwood, 2008). The greater

19
instability of same-sex cohabitation is evident in the plot of the smoothed hazard of dissolution
(Figure 1). Figure 1 also shows that female same-sex cohabitations appear to be slightly more
stable than male same-sex cohabitations, although these figures are based on a small number of
cases, particularly at high durations.
[Table 2, Figure 1]

Differences in the stability of same-sex and different-sex unions

Next, I estimate discrete-time event history models that test for differences in the levels
of stability across union types (Table 3). The model in the first panel suggests that male and
female same-sex cohabitations were less stable than are marriages (Z = 18.2 and 13.4,
respectively). Different-sex cohabitations were also less stable than marriages (Z = 46.7). Wald
tests show that male and female same-sex cohabitations had higher rates of dissolution compared
to different-sex cohabitation (2 (1) = 37.7; p < .001 for men; 2 (1) = 12.8; p < .001 for women).
The differences between same-sex and different-sex unions persisted even after age of union
entry and union history are added in the second model, and socioeconomic status is added in the
third model. In the third model, the dissolution rate for male and female same-sex cohabitation is
7.1 and 5.4 times greater, respectively, than the rate for marriage (p < .001). The difference
between different-sex cohabitation and both male and female same-sex cohabitation continues to
be statistically significant in the third model (for men, 2 (1) = 48.4; p < .001; for women, 2 (1) =
13.3; p < .001). These results suggest that these few demographic characteristics of individuals
and their unions explain little of the variation in stability between same-sex and different-sex
unions. Of course, these models do not control for other potentially confounding factors such as
relationship values and risk aversion, an issue I elaborate on in the Discussion.

20
[Table 3]
Table 3 also tests for differences between male and female same-sex cohabitation. Wald
tests show that there was little difference between female and male couples in the first model (2
(1) = 1.6; p = .20) and second model (2 (1) = 2.4; p = .12). The third model, which introduces
controls for socioeconomic status, suggests that female couples have slightly lower rates of
dissolution than do male couples, though the difference is marginally significant (2 (1) = 2.8; p =
.095). To determine whether my method of classifying same-sex couples biased comparisons in
the stability of same-sex unions, I repeated the analysis using the alternative classification
scheme (Di Salvo, 1995). The results were not substantively different from the results using my
classification of same-sex couples. The models in Table 3 assume that the associations between
the covariates and dissolution are the same for each union type. In the next section, I test this
assumption.

The correlates of dissolution within and between union types

In Table 4, I present the results of discrete-time event history models predicting union
dissolution, stratified by union type. For same-sex cohabitation, the odds ratios for the two spline
functions of duration were 1.02 and 0.99, respectively, and are statistically significant. These
parameters imply that the dissolution rate increases by 2% each month for the first two years of a
union and decreases by 1% by each month thereafter. The associations between duration and
dissolution were similar for same-sex cohabitation and both different-sex cohabitation and
marriage.
[Table 4]

21
There is suggestive evidence that female same-sex cohabitations experience lower rates
of dissolution compared to male same-sex cohabitations. The parameter for gender was not
statistically significant by itself (Z = -1.4) in Table 4, but the Wald test between male and female
same-sex cohabitations in the previous table (Table 3) was marginally statistically significant.
The conclusion about greater instability among male same-sex couples must be regarded
tentatively.
Many of the correlates of dissolution were similar between same-sex and different-sex
unions. There is no evidence of cohort differences in the stability of same-sex cohabitation (Z = 0.2), different-sex cohabitation (Z = 0.7), or marriage (Z = -0.1). Growing up in a non-traditional
family was associated with increased risk of dissolution for all union types, although the
parameter for same-sex cohabitation itself was not statistically significant (Z = 1.1). Like
different-sex cohabitation and marriage, the relationship between age of entry and dissolution for
same-sex cohabitation followed an inverted U-shape.
Consistent with expectations, the relationship between having been in a previous union
and dissolution varies by union type. Having been in a previous union was not associated with
the stability of same-sex cohabitation, but was associated with an increased risk of dissolution
for marriage (O.R. = 1.81; Z = 9.3) and different-sex cohabitation (O.R. = 1.16; Z = 4.1). Wald
tests confirm that this association was significantly stronger for marriage than for same-sex
cohabitation (2 = 4.2; p = .04), but there was no difference between same-sex and different-sex
cohabitation (2 = 0.0; p = .96).
There was no significant association between occupation/employment and dissolution for
same-sex cohabitation. Supplementary analyses showed that this association did not vary by
gender for same-sex cohabitation, but it does for different-sex unions (results not shown). For

22
different-sex cohabitation, not working had a positive association with dissolution for men, but
there was no significant association for women. For marriage, being in a high-skilled rather than
low skilled occupation was negatively associated with dissolution (Z = -2.5) for both men and
women. For men, however, not working was positively associated with marital dissolution, but
was negatively associated with marital dissolution for women.

Discussion
By pooling retrospective data from two birth cohorts, I studied how the levels and
correlates of stability depend on sexual orientation (same-sex versus different-sex unions),
gender (male versus female same-sex unions), and cohort (1958 versus 1970). Three main
conclusions emerge from this study. First, the dissolution rate for male and female same-sex
cohabiters was seven and five times higher, respectively, the rate for marriage. Among
cohabiters, the differences were smaller: the dissolution rate for male and female same-sex
cohabiters was approximately double the rate for different-sex cohabiters. The direction and
magnitude of the differentials are consistent with previous research in other countries (Andersson
et al., 2006; Kalmijn et al., 2007). One potential limitation of using the self-reported NCDS and
BCS data is that same-sex unions might be underreported due to social desirability concerns,
which may bias the group comparisons. Underreporting of same-sex unions likely attenuates
differences between same-sex and different-sex couples. This is because short-term same-sex
unions are more likely to be underreported than longer-term unions: Shorter-term unions are less
salient in peoples lives and easier to conceal in a detailed life history interview. Further, longerterm same-sex cohabiters are also likely to have more experience being open about their
relationships, making them less susceptible to social desirability biases. If short-term same-sex

23
unions are underreported, then observed rates of same-sex dissolution would be downwardly
biasedleading to an underestimate of differences between same-sex and different-sex unions.
The findings are consistent with the perspective the lack of legal and social
institutionalization of same-sex couples may lead same-sex couples to perceive fewer rewards,
fewer barriers, and more alternatives to their unionsleading to higher rates of dissolution. This
perspective is grounded in longstanding theory about institutionalization and investments and
consistent with previous research on how same-sex couples organize their relationships. A
competing explanation, however, is that the elevated rate of dissolution is due to the types of
people who enter same-sex unions. If individuals who choose to cohabit with a same-sex partner
have more liberal attitudes toward dissolution or are more open to change and risk taking, then
higher dissolution rates may stem from selection processes rather than the lack of marriage and
the normative climate. There is some evidence to support this reasoning. Although all sexual
orientation groups value commitment and faithfulness, sexual orientation minorities are less
likely to view these values as ingredients to a successful relationship (Meier, Hull, & Ortyl,
2009). These values may stem from the fact that same-sex couples may have different goals in
pursuing relationships: while heterosexual life is largely grounded in marriage and childbearing,
the lack of marriage for same-sex couples and greater difficulty having children means that there
is not a logical endpoint for same-sex relationships (Strohm et al., 2009). Different relationship
values, however, are insufficient to fully explain the results in this study: although different-sex
cohabiters share many of the same individualistic values as same-sex couples (Clarkberg,
Stolzenberg, & Waite, 1995), same-sex cohabiters still had higher dissolution rates compared to
different-sex cohabiters. It is likely that both institutional and selection processes are at work.

24
Future research that integrates attitudes and demographic behavior will be able to shed light on
the relative importance of the selection and institutional perspectives.
Second, there was some evidence that male same-sex couples were less stable than were
female couples. Although this finding should be interpreted as suggestive due to its marginal
significance, this result may help to resolve the mixed research on the relative stability of male
and female same-sex couples. My finding is consistent with another study that observed greater
instability among male same-sex cohabiters in the Netherlands (Kalmijn et al., 2007). Another
study, however, of same-sex married couples in Norway and Sweden found that female couples
were less stable (Andersson et al., 2006). What explains these discrepant findings? One potential
explanation concerns gender differences in the types of couples who choose to marry. Same-sex
couples who chose to marry are more committed and have been together for longer than
unmarried couples (Fingerhut & Maisel, 2010). But this selection process may be stronger for
men than for women: a broad range of female couples may marry, but only the most committed
male couples may marry. Indeed, male couples who chose to legally recognize their unions have
been together longer than their female counterparts (Carpenter and Gates, 2008). Thus, male
same-sex married couples in the 2006 study by Andersson et al. may have been particularly
committed to their relationship compared to female couples, resulting in greater stability for male
couples. To test this hypothesis, more descriptive studies of same-sex marriages and
cohabitations in diverse contexts are needed. These studies could pave the way for research on
how womens greater social psychological investment in relationships and different economic
circumstances affect the stability of relationships more generally.
Third, consistent with previous research (Andersson et al., 2006; Kalmijn et al., 2007),
there were few differences in the correlates associated with the stability of same-sex and

25
different-sex unions. A notable exception was previous cohabitation or marriage, which was
positively associated with dissolution for marriage, but had no association for same-sex
cohabitation. Future research is needed to determine whether this null finding reflects true
behavior or a lack of statistical power. My prediction of cohort differences in the stability of
same-sex unions received no support. This may be due to three factors. First, the twelve year
difference in birth cohorts may not be long enough to observe the effects of a changing
normative climate. Second, small sample sizes also might not have provided sufficient statistical
power to detect a difference. Third, the effects of the increasingly tolerant social climate (leading
to greater stability in the 1970 cohort) may have been offset by selection processes, whereby
only especially committed couples from the 1958 cohort chose to live together or report their
relationship on a survey (leading to greater stability in the 1958 cohort). The plausibility of this
selection process suggests that researchers studying cohort changes in same-sex couples should
account for changes in the types of people who enter a same-sex union, possibly by measuring
relationship commitment at the beginning of a panel study (Balsam et al., 2008) or by measuring
attitudes or values in a cross-sectional study.
This study complements previous demographic studies on the stability of same-sex
unions (Andersson et al., 2006; Kalmijn et al., 2007) and the burgeoning demographic literature
on same-sex couples more generally. The limitations of this study described aboveits reliance
on self-report data, small samples, and lack of data on attitudes and valuespoint to directions
for future research and the need for more data on same-sex couples. The investigation of gender
and cohort differences in same-sex union stability also underscores the importance of
incorporating selection effects into demographic studies of same-sex couples. By addressing
these methodological challenges, researchers will be able to exploit the unique social context of

26
same-sex couples to study how institutionalization, gender, and social change affect intimate
unions more generally (Blumstein & Schwartz, 1983).

References
Andersson, G., Noack, T., Seierstad, A., and Weedon-Fekjaer, H. (2006). The demographics of
same-sex marriages in Norway and Sweden. Demography, 43, 79-98.
Balsam, K. F., Beauchaine, T. P., Rothblum, E. D., and Solomon, S. E. (2008). Three-year
follow-up of same-sex couples who had civil unions in Vermont, same-sex couples not in
civil unions, and heterosexual married couples. Developmental Psychology, 44, 102-116.
Berrington, A. (2003). Change and continuity in family formation among young adults in
Britain. Applications and Policy Working Paper A03/04. Southampton Statistical
Sciences Research Institute, Southampton, UK.
Berrington, A. and Diamond, I. (1999). Marital dissolution among the 1958 British birth cohort:
the role of cohabitation. Population Studies, 53, 19-38.
Biblarz, T. J. & Savci, E. (2010). Lesbian, gay, bisexual, and transgender families. Journal of
Marriage and Family, 72, 480-497.
Black, D. A., Gates, G. J., Sanders, S. G., and Taylor, L. J. (2000). Demographics of the gay and
lesbian population in the United States: evidence from available systematic data sources.
Demography, 27, 139-154.
Black D. A., Sanders, S. G., & Taylor, L. J. (2007). The economics of lesbian and gay families.
Journal of Economic Perspectives, 21, 53-70.
Blumstein, P. & Schwartz, P. (1983). American couples: money, work, sex. New York: William
Morrow.

27
Brines, J. & Joyner, K. (1999). The ties that bind: principles of cohesion in cohabitation and
marriage. American Sociological Review, 64, 333-355.
Bynner, J., Butler, N., Ferri, E., Shepherd, P, and Smith, K. (2005). The design and conduct of
the 1999-2000 surveys of the National Child Development Study and the 1970 British
Cohort Study. Centre for Longitudinal Studies Working Paper 1. Centre for Longitudinal
Studies, Institute of Education, University of London.
Carpenter, C. & Gates, G. J. (2008). Gay and lesbian partnership: evidence from California.
Demography, 45, 573-90.
Cherlin, A. J. (2004). The deinstitutionalization of American marriage. Journal of Marriage and
Family, 66, 848-861.
Clarkberg, M., Stolzenberg, R. M., & Waite, L. J. (1995). Attitudes, values, and entrance into
cohabitational versus marital unions. Social Forces, 74, 609-632.
Cross, S. E. & Madson, L. (1997). Models of the self: self-construals and gender. Psychological
Bulletin, 122, 5-37.
Di Salvo, P. (1995). NCDS5 partnership histories. National Child Development Study Data Note
2. Center for Longitudinal Studies, Institute of Education, Centre for Longitudinal
Studies, University of London.
Ellingson, S. & Schroeder, K. (2004). Race and the construction of same-sex sex markets in four
Chicago neighborhoods. In E. O. Laumann, S. Ellingson, J. Mahay, A. Paik, & Y. Youm.
(Eds.), The sexual organization of the city (pp. 93 126). Chicago: The University of
Chicago Press.

28
England, P. & Kilbourne, B. S. (1990). Markets, marriages, and other mates: the problem of
power. In R. Friedland & A. F. Robertson (Eds.), Beyond the marketplace: rethinking
economy and society (pp. 163-188). New York: Aldine de Gruyter.
Ermisch, J. and Francesconi, M. (2000). Cohabitation in Great Britain: not for long, but here to
stay. Journal of the Royal Statistical Society Series A, 163, 153-171.
European Commission (2007). Eurobarometer 66: Public opinion in the European Union.
Retrieved from http://ec.europa.eu/public_opinion/archives/eb/eb66/eb66_en.htm
Fingerhut, A. W. & Maisel, N. C. (2010). Relationship formalization and individual and
relationship well-being among same-sex couples. Journal of Social and Personal
Relationships, 27, 956-969.
Gates, G. J. & Steinberger, M. D. (2010). Same-sex unmarried partner couples in the American
Community Survey: the role of misreporting, miscoding and misallocation. Unpublished
manuscript, Pomona College Department of Economics.
http://economics-files.pomona.edu/steinberger/index_files/page0001.htm
Gregory, M., Zissimos, B., and Greenhalgh, C. (2001). Jobs for the skilled: how technology,
trade, and domestic demand changed the structure of UK employment, 1979-1990.
Oxford Economic Papers, 53, 20-46.
Guimond, S., Chatard, A., Martinot, D., Crisp, R., & Redersdorff, S. (2006). Social comparison,
self-stereotyping, and gender differences in self-construals. Journal of Personality and
Social Psychology, 90, 221-242.
Haskey, J. (2001). Cohabitation in Great Britain: past, present, and future trends and attitudes.
Population Trends, 103, 4-25.

29
Hawkes, D. and Plewis, I. (2006). Modelling non-response in the National Child Development
Study. Journal of the Royal Statistical Society, Series A, 169, 479-491.
Heimdal, K.R. & Houseknecht, S. K. (2003). Cohabiting and married couples income
organization: approaches in Sweden and the United States. Journal of Marriage and
Family, 65, 525-538.
Herek, G. M. (2006). Legal recognition of same-sex relationships in the United States: a social
science perspective. American Psychologist, 61, 607-621.
Jivani, A. (1997). Its not unusual: A history of lesbian and gay Britain in the twentieth century.
Bloomington: Indiana University Press.
Kalmijn, M., Loeve, A., & Manting, D. (2007). Income dynamics in couples and the dissolution
of marriage and cohabitation. Demography, 44, 159-179.
Kroester, C. & Booth, A. (2000). Barriers to divorce: when are they effective? When are they
not? Journal of Family Issues, 21, 78-99.
Kurdek, L. A. (1995). Assessing multiple determinants of relationship commitment in cohabiting
gay, cohabiting lesbian, dating heterosexual, and married heterosexual couples. Family
Relations, 44, 261-266.
Kurdek, L. A. (1998). Relationship outcomes and their predictors: longitudinal evidence from
heterosexual married, gay cohabiting, and lesbian cohabiting couples. Journal of
Marriage and the Family, 60, 553-568.
Kurdek, L. A. (2003). Differences between gay and lesbian cohabiting couples. Journal of Social
and Personal Relationships, 20, 411-436.
Kurdek, L. A. (2004). Are gay and lesbian cohabiting couples really different from heterosexual
married couples? Journal of Marriage and Family, 66, 880-900.

30
Lehmiller, J. J. & Agnew, C. R. (2006). Marginalized relationships: the impact of social
disapproval on romantic relationship commitment. Personality and Social Psychology
Bulletin, 32, 40-51.
Lehmiller, J. J. & Agnew, C. R. (2007). Perceived marginalization and the prediction of romantic
relationship stability. Journal of Marriage and Family, 69, 1036-1049.
Levinger, G. (1965). Marital cohesiveness and dissolution: an integrative review. Journal of
Marriage and Family, 27, 19-28.
Levinger, G. (1976). A socio-psychological perspective on marital dissolution. Journal of
Social Issues,52, 21-47.
Meier, A., Hull, K. H., & Ortyl, T. A. (2009). Young adult relationship values at the intersection
of gender and sexuality. Journal of Marriage and Family, 71, 510-525.
Meyer, I. (2003). Prejudice, social stress, and mental health in lesbian, gay, and bisexual
populations: conceptual issues and research evidence. Psychological Bulletin, 129, 674697.
Muraco, A. (2006). Intentional families: fictive kin ties between cross-gender, different sexual
orientation friends. Journal of Marriage and Family, 68, 1313-1325.
Previti, D. & Amato, P. R. (2003). Why stay married? Rewards, barriers, and marital stability.
Journal of Marriage and Family, 65, 561-573.
Solomon, S. E., Rothblum, E. D., & Balsam, K. F. (2004). Pioneers in partnership: lesbian and
gay male couples in civil unions compared with those not in civil unions and married
heterosexual siblings. Journal of Family Psychology, 18, 275-286.
Steele, F., Kallis, C. Goldstein, H. & Joshi, H. (2005). The relationship between childbearing and
transitions from marriage and cohabitation in Britain. Demography, 42, 647-673.

31
Steele, F., Kallis, C., & Joshi, H. (2006). The formation and outcomes of cohabiting and marital
partnerships in early adulthood: the role of previous partnership experience. Journal of
the Royal Statistical Society, Series A, 169, 757-779.
Strohm, C. Q., Seltzer, J. A., Cochran, S. D., and Mays, V. M. (2009). Living apart together
relationships in the United States. Demographic Research, 21, 177-214.
Tobin, B. (2009). Same-sex couples and the law: recent developments in the British Isles.
International Journal of Law, Policy, and the Family, 23, 309-330.
Treas, J. (1993). Money in the bank: transaction costs and the economic organization of
marriage. American Sociological Review, 58, 723-734.
Weston, K. (1997). Families we choose: lesbians, gays, kinship. New York: Columbia
University Press.
Wilson, B. & Smallwood, S. (2008). The proportion of marriages ending in divorce. Population
Trends, 131, 28-36.

32
Tables
Table 1. Characteristics of Individuals (N = 20,070) and Their Unions (N = 25,656)
Ever in
Ever in
A. Characteristics of Individuals
same-sex
different-sex
cohabitation
cohabitation
Ever married
(n = 186
(n = 13,266
(n = 15,649
individuals)
individuals)
individuals)
Female
47
51
54
Cohort
1958
27
36
59
1970
73
64
41
Region of residence in childhood
London and Southeast
30
25
21
Scotland and the North
27
35
38
Wales and the Midlands
26
22
23
South and East
15
16
18
Missing or abroad
2
2
1
Family disruption in childhood
31
32
25
Mothers years of postcompulsory education
0
57
58
57
1-2
27
25
28
3 or higher
9
11
9
Missing
7
6
7
1
Enrolled in full-time education
14
7
6
Missing data on enrollment 1
5
6
8
1
Occupation
High-skilled
44
33
30
Medium-skilled
34
36
37
Low-skilled
5
11
11
Not working
12
14
16
Missing
5
6
6
B. Characteristics of Unions
Same-sex
Different-sex
cohabitation
cohabitation
Marriages
(n = 263
(n = 17,219
(n = 16,837
unions)
unions)
unions)
Number of previous unions
Zero
60
71
95
One
28
24
5
Two or more
12
5
<1
Median age at union entry (years)
25.9
24.6
24.5
Note: Variables defined in text. Totals may not sum to 100 due to rounding. Percentages
Unless Noted.
1
Time-varying variable. Education enrollment is shown at age 23; occupation is shown at
age 29.

33
Table 2. Survival Probabilities by Union Type and Duration (N = 25,656 unions)
Same-Sex Cohabitation

Different-Sex

All

Men

Women

Cohabitation1

263

138

125

17,219

8,174

.79

.76

.83

.90

.98

.67

.65

.69

.82

.95

.54

.52

.55

.76

.93

.47

.46

.48

.71

.91

.37

.37

.38

.67

.88

.33

.29

.38

.64

.86

.28

.24

.34

.62

.84

.25

.21

.31

.60

.82

# of unions

Marriage

Duration (years)

Includes unions that transitioned to marriage and those that did not.

34

Table 3. Odds Ratios (OR) From a Discrete-Time Event History Analysis of Dissolution (N =
25,656 unions)
OR

OR

OR

1.01
0.99

7.1
-13.6

1.01
0.99

7.5
-17.4

1.01
0.99

8.0
-16.9

6.93
5.62
3.52

18.2
13.4
46.7

7.19
5.57
3.47

18.5
13.2
45.2

7.12
5.43
3.42

18.5
13.2
44.7

1970 cohort (1958 cohort)

Childhood region (Southeast, South, East)
Scotland, North, Wales, Midlands
Missing

0.94

-2.3

0.97

-1.3

0.99

-0.3

0.90
1.02

-4.6
0.2

0.90
1.02

-4.4
0.2

0.90
1.02

-4.5
0.2

Non-traditional family in childhood

Mothers education (0-2 years)1
3 or more
Missing

1.28

10.3

1.20

7.7

1.20

7.6

1.15
1.10

3.8
2.0

1.22
1.14

5.2
2.7

1.20
1.13

4.7
2.6

Age entered union (years)

Age entered union2 (years)

0.72
1.01

-9.7
7.9

0.73
1.01

-9.3
7.6

Had previous union(s) (none)

Enrollment in education (not enrolled)
Enrolled
Missing enrollment

1.29

7.9

1.29

7.8

1.48
1.27

6.5
3.6

0.93
0.96
0.98
1.12

-1.7
-1.0
-0.5
2.2

Duration
Spline years 1-2
Spline years 3+
Union type (marriage)
Male same-sex cohabitation
Female same-sex cohabitation
Different-sex cohabitation

Occupation (low-skilled)
High-skilled
Medium-skilled
Not working
Missing
Log-likelihood
-52,624
-52,439
Note: Variables described in text. Reference category is in parentheses.
1
Mothers years of post-compulsory schooling.

-52,404

35

Table 4. Odds Ratios (OR) From a Discrete-Time Event History Analysis of Dissolution,
Stratified by Union Type (N = 25,656 unions)
Same-Sex
Different-Sex
Cohabitation
Cohabitation
Marriage
Duration
OR
Z
OR
Z
OR
Z
Spline years 1-2
1.02
2.4
1.01
6.7
1.04
8.9
Spline years 3+
0.99
-3.0
0.99 -14.0
0.99 -13.1
Female

0.80

-1.4

0.83

-6.0

0.92

-2.1

1970 cohort (1958 cohort)

Childhood region (SE, South, East)
Scotland, North, Wales, Midlands
Missing

0.96

-0.2

1.02

0.7

1.00

-0.1

0.83
0.55

-1.2
-1.0

0.88
0.96

-4.1
-0.4

0.93
1.12

-2.0
0.6

Non-traditional family in childhood

Mothers education (0-2 years) 1
3 or more years
Missing

1.20

1.1

1.14

4.3

1.31

6.9

0.83
1.62

-0.6
2.1

1.30
1.21

5.9
3.1

0.97
1.04

-0.4
0.4

Age entered union (years)

Age entered union2 (years)

0.69
1.01

-1.7
1.6

0.87
1.00

-3.3
2.5

0.65
1.01

-6.3
4.3

Had previous union(s) (none)

Enrolled in education (not enrolled)
Enrolled
Missing enrollment
Occupation (low-skilled)
High-skilled
Medium-skilled
Not working
Missing

1.18

0.8

1.16

4.1

1.81

9.3

1.15
1.75

0.3
1.6

1.43
1.18

5.3
1.8

0.96
1.34

-0.2
3.3

1.37
1.20
1.07
1.44

0.7
0.4
0.1
0.7

1.03
1.00
1.22
1.19

0.6
0.1
3.4
2.4

0.85
1.00
0.82
1.13

-2.5
0.1
-2.9
1.5

Log-likelihood
-841
-29,240
Number of couples
263
17,219
Note: Variables described in text. Reference category is in parentheses.
1
Mothers years of post-compulsory schooling.

-22,125
16,837

36

Figure

.015
.01
0

.005

Hazard (smoothed)

.02

Figure 1: Rate of Dissolution by Union Type

12

24
36
Duration (months)

Male Same-Sex
Different-Sex Cohab.

48
Female Same-Sex
Marriage

60